Cytotoxic CD8 T cells are key effectors in the immunotherapy of malignant and viral diseases. However, the lack of efficient methods for their in vitro priming and expansion has become a bottleneck to the development of vaccines and adoptive transfer strategies. Synthetic artificial APCs (aAPCs) are now emerging as an attractive tool for eliciting and expanding CTL responses. We show that, by controlling the MHC density on aAPCs, high- or low-avidity tumor-directed human CTL lines can be raised effectively in vitro if costimulation via CD28 and IL-12 is provided. Compared with low-avidity CTL lines, high-avidity CTLs need 100- to 1000-fold less peptide for activation, bind more MHC tetramers, and, as expected, are superior in recognizing tumor cell lines expressing Ag. We believe that the possibility to raise Ag-specific T cells with predetermined avidity will be crucial for the future use of aAPCs in immunotherapeutical settings.
In the host defense against viral infections, CD8 T cells play a central role, and these cells are thought to be key effectors in immunotherapy of malignant diseases. There are two main reasons to elicit and expand CTLs in vitro. First, following reverse immunology, candidate T cell epitopes are first identified, and their immunogenicity is validated subsequently (1). Second, for adoptive transfer, large numbers of antiviral or antitumor CTLs are expanded ex vivo (2). However, the currently applied approaches to elicit primary in vitro CTL responses usually involve differentiation of dendritic cells (DCs) 3 from the patient’s blood, which is time-consuming, expensive, and limited by the obtainable cell numbers.
Therefore, it would be highly desirable to have artificial APCs (aAPCs) at one’s disposal for in vitro experiments. Although some studies have used tumor cell lines transfected with MHC:peptide complexes and costimulatory molecules for this purpose (3), the most rigorously controlled aAPC would involve coating of synthetic surfaces, usually cell-sized plastic microspheres, with purified MHC plus costimulatory molecules. This is basically a very old idea (4), and several groups have relied on it to investigate the events that lead to the activation of T cells, mostly in the mouse system (5, 6, 7). Soluble MHC reagents were also used for this purpose (5), but found to be less potent when directly compared with immobilized MHC. However, it was only very recently (8) that aAPCs were shown to be able to fully induce in vitro priming of human CD8 T cells and to sustain long-term cellular proliferation, as required for any immunotherapeutical approach. However, this last study used empty HLA-Ig fusion proteins coated on microspheres and subsequent peptide loading. The efficiency of this process, and thus the functional Ag density on aAPCs, will strongly depend on the affinity of an individual peptide to the MHC molecule. This precluded an easy control of the number of MHC:peptide complexes on the aAPC. However, it is known that the Ag dose will influence the avidity of the responding T cell population (9), with high-avidity CTLs being superior in adoptive transfer experiments.
In this study, we rely on preformed MHC:peptide complexes coupled by biotin:streptavidin biochemistry to the surface. This system allows the exact control of the MHC density on aAPCs, which enables us to selectively elicit high- or low-avidity Ag-specific CTL responses with high efficiency from healthy human individuals. Furthermore, we identify the factors that are necessary for successful in vitro priming with aAPCs.
Materials and Methods
Peptides, recombinant MHC molecules, fluorescent tetramers, and MHC-coated microspheres
Peptides in this study (Ref.10 ; http://www.syfpeithi.de/) were synthesized using standard Fmoc chemistry. The peptide library consisted of an approximately equimolar mixture of 6912 nonapeptides with the structure Y, L/M, A/I/L/Y, E/G/P, G/K/L, I/V/L, A/V/P/H, E/K/S/T, and V/L. Biotinylated recombinant HLA-A*0201 (A*02) molecules and fluorescent MHC tetramers were produced as described previously (11). The costimulatory mouse IgG2a anti-human CD28 Ab 9.3 (12) was biotinylated using sulfo-N-hydroxysuccinimidobiotin as recommended by the manufacturer (Perbio Science, Bonn, Germany). As a negative control, biotinylated mouse IgG2a Ab G155-178 (BD Biosciences, Heidelberg, Germany) was used. For generation of aAPCs, 5.6-μm-diameter streptavidin-coated polystyrene particles with a binding capacity of 0.064 μg of biotin-FITC per milligram of microsphere (Bangs Laboratories, Fishers, IL) were resuspended at 2 × 106 particles per milliliter in buffer containing biotinylated MHC and Abs at indicated concentrations and incubated at room temperature for 30 min.
Ag-specific in vitro stimulation of human CD8 T cells
PBMCs were isolated from fresh buffy coats using standard gradient separation. When indicated, untouched CD8 T cells were MACS enriched by negative depletion (Miltenyi Biotec, Bergisch Gladbach, Germany).
To compare DC to bead stimulations, monocyte-derived human DCs were generated as previously described (132 (Sigma-Aldrich, Taufkirchen, Germany) for 2–3 days. Mature DCs were predominantly CD14−CD40+CD80+CD83+CD86+ and HLA-DRhigh (data not shown). For restimulations after priming with DCs, cryopreserved autologous PBMCs were used.
In vitro stimulations were initiated in 24-well plates with 5 × 106 responder cells plus 1 × 106 beads or 1 × 106l
Ag-specific T cell enrichment and expansion
FACS sorting of bead-stimulated cells was performed on a FACSVantage after staining with PE tetramers and Abs CD8-allophycocyanin clone SK1 and CD4-FITC (BD Biosciences).
Alternatively, bead-stimulated cells were restimulated with irradiated T2 cells pulsed with 5 μM peptide as described above, and IFN-γ+ cells were enriched by MACS (Miltenyi Biotec).
Sorted cells were cultured in the presence of 5 × 105 cells/ml irradiated fresh allogenic PBMCs, 5 × 104 cells/ml irradiated LG2-EBV cells, 150 U/ml IL-2, and 0.5 μg/ml PHA-L (Roche Diagnostics, Mannheim, Germany). Cells were further expanded in T cell medium containing 150 U/ml IL-2.
Cell surface/intracellular cytometric analysis
Tetrameric analyses were performed with fluorescent MHC tetramers plus Abs CD4-FITC and CD8-PerCP clone SK1 on a four-color FACSCalibur (BD Biosciences). Total specific cell numbers per sample were calculated by FACS analysis as follows after adding a defined number of microspheres to each sample: (specific cells counted) × (microspheres added)/(microspheres counted).
Intracellular cytometry was performed using a Cytofix/Cytoperm Plus kit with Abs CD4-FITC, IFN-γ-PE, and CD8-PerCP, and analyzed on a FACSCalibur cytometer (BD Biosciences).
Cytotoxicity was tested in a standard 4-h 51Cr release assay using 3000 target cells per well. Percentage of specific lysis was calculated as follows: (experimental release − spontaneous release)/(total release − spontaneous release) × 100.
Results and Discussion
MHC/anti-CD28-coated microspheres are powerful tools for in vitro priming
Streptavidin-linked 5.6-μm polystyrene microspheres could be easily coated with biotinylated MHC molecules and costimulatory anti-CD28 Ab. Moreover, labeling, as indicated by immunofluorescence, remained almost constant over a storage period of 4 wk (data not shown). Therefore, such particles have immense practical advantages over the use of DCs, which can only be generated in limited amounts in a time-consuming manner.
To test the capacity of beads as APCs, human CD8 T cells were stimulated for three 7- to 9-day rounds in the presence of IL-12 with beads coated with anti-CD28 Ab- plus 10 nM A*02-bound epitopes derived from a viral Ag (CMV pp65), a modified self Ag (Melan-A), or a tumor-derived self Ag (MET proto-oncogene). As determined by tetramer analysis, stimulation with beads led in all cases to a specific CTL expansion with the correspondent specificity (Fig. 1⇓, middle panel). There was no staining with an irrelevant tetramer (Fig. 1⇓, right panel). When stimulated with beads containing irrelevant MHC molecules, a specific tetramer+ population was visible only in the case of the virus recall Ag (Fig. 1⇓, left panel; the donor in this experiment was human CMV seropositive). The stimulation with the modified self Ag from Melan-A was especially efficient, which was also confirmed in experiments with PBMCs from different donors (later figures and data not shown). This is consistent with the common observation of relatively high precursor frequencies in the blood of healthy A*02+ donors against this peptide. However, as these cells appear to be naive (14) and require professional APCs to be expanded in vitro (15), our findings indicate that beads were capable of efficient in vitro priming.
Microsphere-expanded CTLs are functional
The expanded T cells were functional, because they specifically expressed IFN-γ upon restimulation with peptide, as indicated by intracellular cytokine staining. Interestingly, this was also the case when beads were used for CTL priming that were coated with an A*02/peptide library, indicating that aAPCs may also be useful for stimulating with complex Ag mixtures (data not shown).
To investigate the cytotoxic capabilities of expanded CD8 T cells, fractions from T cell lines were sorted and subsequently expanded using mitogen and IL-2. T cells proliferated strongly for at least 3 mo under such conditions (data not shown). Specific cytotoxicity was shown for all tested CTLs and was confined to the tetramer+ or IFN-γ+ fraction (Fig. 2⇓).
CTL priming by low MHC density aAPCs requires costimulation and IL-12
These data indicated that bead-expanded CTLs were Ag specific and functional. However, several of these CTL lines did not recognize target cells endogenously processing Ag (data not shown). As an explanation, we speculated that beads coated with 10 nM MHC (high-density beads) as used for the above experiments may lead to the preferential expansion of low-avidity CTLs. To test this hypothesis, we titrated the specific MHC molecules during bead coating. Efficient expansions were still observed using 100-fold less of the specific MHC molecules (low-density beads) than for above experiments (Fig. 3⇓A, left plot). The results were similar when total MHC concentrations were kept constant by adding an MHC library (Fig. 3⇓A, right plot) as a filler.
All of the stimulations described above were performed in the presence of CD28 Ab on aAPCs and exogenous IL-12. To identify obligatory factors for priming, we varied stimulation conditions (Fig. 3⇑B). Using high-density beads, T cell lines could be generated even in the absence of costimulatory CD28 Ab or exogenous IL-12, although efficiency was greatly reduced when both were missing. However, for low-density beads, the presence of CD28 Ab as well as exogenous IL-12 was obligatory for successful in vitro priming. To compare efficiency with a well-established protocol, this was done in parallel with one stimulation by peptide-pulsed autologous mature DCs and restimulations by autologous peptide-pulsed PBMCs. Compared with stimulations by DCs/PBMCs, MHC-coated high- or low-density beads were at least as efficient or even superior, especially in terms of total numbers of specific cells present. Exogenous IL-12 p70 also enhanced CTL responses initiated by DCs plus PBMCs, which is well in accordance with previous studies describing importance of IL-12 p70 in CD8 T cell priming (16). These data support a three-signal model for the priming of human CD8 T cells that has been suggested previously for mice (7). Nevertheless, signals 2 and 3 (ligation of costimulatory receptors such as CD28 and inflammatory cytokines such as IL-12) may become redundant in the presence of an extremely strong signal 1, as shown previously for mice (6).
During tetramer analysis, we noted that CTLs generated by low-density beads bound higher amounts of MHC tetramers (Fig. 3⇑C). According to some previous studies, this could indicate higher avidity of CTLs (17), although this may not always be the case (18).
MHC density during priming influences avidity of resulting CTLs
To clarify the question of the avidity of CTLs primed by high- or low-density beads, the amount of antigenic peptide was titrated on target cells in a standard 51Cr release assay (Fig. 4⇓A). The T cell line generated with high-density beads needed ∼10 nM peptide for recognition, whereas the line generated with low-density beads was of much higher avidity and needed only picomolar amounts of peptide for an efficient recognition. Therefore, the Ag density of aAPCs influenced the overall avidity of the responding T cell population. As we used filler MHC molecules on the bead surface to ensure a constant overall MHC density, this effect is likely due to less cognate Ag recognition and not due to less CD8 binding. The most ready explanation for this finding is that the naive CD8 T cell precursor pool consisted of different clones with a broad spectrum of avidities. High-avidity clones are likely to be rarer for a given MHC:peptide complex. When stimulating with a low determinant density, only the high-avidity clones will proliferate. A high Ag dose will lead to the stimulation of many low- and few high-avidity clones. Furthermore, high-avidity clones may be overtly stimulated by high Ag densities and thus die due to exhaustion effects. Overall, these processes would explain the observed results.
Finally, we tested whether bead-generated T cell lines could also recognize tumor cells expressing Melan-A and A*02 (Fig. 4⇑, B–G). One such tumor cell line was solely recognized (Fig. 4⇑D), and one other cell line was much better recognized (F) by T cells primed with low-density aAPCs. Recognition was specific, as shown by an irrelevant T cell line from the same donor (Fig. 4⇑, B–G) and an irrelevant Melan-A−A*02+ target cell line (A).
By controlling the surface density of MHC molecules on coated microspheres, we were able to prime at will high- or low-avidity tumor-directed human CD8 T cells in vitro. Because our method is highly efficient, it could be useful for clinical tumor immunotherapy. MHC-coated microspheres provide the most rigorously controlled Ag-presenting “cells” available and are a powerful tool that allows new insights in key parameters necessary for effective T cell responses.
All melanoma cell lines were a generous gift of Prof. Dr. Claus Garbe (Department of Dermatology, University of Tübingen). We thank Patricia Hrstic and Beate Pömmerl for expert technical assistance.
↵1 This work was supported by grants from the Deutsche Forschungsgemeinschaft (SFB510; C7) and the European Union (QLK2-CT-2002-0062 EPI-PEP-VAC). S.W. is a fellow of the Reinhold Beitlich Stiftung (Tübingen, Germany).
↵2 Address correspondence and reprint requests to Dr. Stefan Stevanović, Department of Immunology, Institute for Cell Biology, Auf der Morgenstelle 15, D-72076 Tübingen, Germany. E-mail address:
↵3 Abbreviations used in this paper: DC, dendritic cell; aAPC, artificial APC; A*02, HLA-A*0201.
- Received July 16, 2003.
- Accepted September 29, 2003.
- Copyright © 2003 by The American Association of Immunologists